The OECD Metropolitan Governance Survey A QUANTITATIVE DESCRIPTION OF GOVERNANCE STRUCTURES IN LARGE URBAN AGGLOMERATIONS

Even though metropolitan areas account for half of the population, and an even larger share of economic activity of OECD countries, almost no systematic information on governance structures in these areas exists. This study – based on a novel data set – gives an overview of governance arrangements in OECD metropolitan areas. It shows that organisations dedicated to metropolitan area governance are common, but often have little powers. Nevertheless, the existence of such organisations is related with better performance on a range of important outcome variables, such as public transport systems, environmental issues, and urban sprawl. JEL classification: H11, H70, R50.


INTRODUCTION
Metropolitan areas face distinct challenges. Land is scarce and subject to competing interests. Complex transportation infrastructures must be built and maintained to avoid congestion. Local income disparities are larger than those found in other parts of a country and create tensions between poor and wealthy neighbourhoods. Public services need to be efficiently provided for millions of people. While these challenges concern a wide range of policy fields, they all have in common that they usually reach beyond the limits of individual municipalities. They require solutions that encompass the entire metropolitan area. However, traditional structures of local governance often do not correspond to the geographical extent of the challenges. Individually, each municipality (or the corresponding lowest level of local governments) is too small to provide solutions to metropolitan area wide problems.
To address this issue, a wide range of solutions to the challenge of metropolitan area governance has been developed. Some of them complement existing local governments and some have replaced old local government structures. They are often shaped by local actors from within the metropolitan area and can differ strongly across metropolitan areas. Although they form an important part of governance arrangements in metropolitan areas, little systematic information is available on their design and even on their existence. This paper presents a new data set on organisations that focus on metropolitan area governance. 3 Such organisations are common in most OECD countries, but typically have few powers and responsibilities. While around two thirds of all metropolitan areas are covered by an organisation that is responsible for its governance, only a quarter of them have the power to impose any legally binding laws or regulations. Corresponding to the few responsibilities that they have, small budgets of less than 30 US dollars (USD) are prevalent among most organisations. Nevertheless, those organisations that have more powers can have annual per capita budgets of several thousand USD.
Organisations responsible for metropolitan governance work on a wide range of topics. However, three fields of work are significantly more common than others. These are regional economic development, spatial planning and transportation. Although most organisations have few powers and small budgets, correlations between their existence and desirable outcomes can be found in several fields. Where organisations responsible for metropolitan governance exist, metropolitan areas tend to be larger but record lower levels of urban sprawl. Similarly, the existence of organisations that focus on the provision of public transport for the entire metropolitan area is correlated to higher levels of public satisfaction with public transport provision and lower levels of air pollution.

3.
The data used in this paper is provided in Appendix 2. It can be obtained in digital form from the corresponding author Abel Schumann (abel.schumann@oecd.org).

Background and Motivation
No agreement on the optimal governance structure of metropolitan areas exists in the literature. Prominent proponents of public choice theories, such as Tiebout (1956), V. Ostrom et al. (1961), E. Ostrom et al. (1978) and Parks and Oakerson (1989) support polycentric or fragmented governance arrangements. E. Ostrom (2010) mentions several main arguments that the literature provides in favour of polycentric governance structures. Smaller local governments are better in representing preferences of citizens and allow for more participation of individuals. They also tend to be better in monitoring performance and costs of service provision. Furthermore, smaller jurisdictions make it easier for citizens to move to a municipality that offers a desired mix of taxes and service provision.
Proponents of these arguments argue that even in metropolitan areas with highly fragmented governance structures, informal coordination mechanisms between local authorities exist that ensure effective delivery of public services at different geographical scales. There is some empirical support for this school of thought. For example, Carr and Feiock (1999), Bish (2001), Moisio and Uusitalo (2006), Martin andHock Schiff (2011), or Aulich, Sansom andMckinlay (2013) analyse consolidations of local governments, which reduce the complexity of governance structures in metropolitan areas. They do not find that these consolidations have positive effects on public sector efficiency or economic growth. However, as these studies generally do not take underlying differences between municipalities that consolidated and those that did not consolidated into account, their informative value is limited.
In contrast to the previous theoretical and empirical studies that broadly support polycentric government structures in metropolitan areas, other inquiries into the subject come to the opposite conclusion. Centrist approaches to local governance have existed for many decades and in fact the development of public choice approaches to local governance can be seen as a response to them. Authors such as Peirce et al. (1993), Orfield (1997) and Savitch and Vogel (2000) have argued that urban agglomerations have outgrown the old administrative boundaries of municipalities. Consequently, governance structures have to be adapted to the new realities. This literature claims that urban sprawl and regional inequality are fostered by administrative fragmentation and that socially harmful competition for economic development is likely to occur among municipalities. Furthermore, the absence of coordination hinders large-scale metropolitan-wide infrastructure projects. Analysing the English context, Norris (2001) finds that metropolitan wide governance virtually ceased to exist with the abolishment of the Metropolitan County Councils in the 1980s. In a broader empirical study of American metropolitan areas, Nelson and Foster (2002) find that less fragmentation is positively correlated to per capita income growth. Building on these results, Stansel (2005) finds that the existence of general-purpose governments (i.e. counties) is positively and significantly correlated with population growth and per capita income growth. Throughout all studies, however, it remains unclear whether the correlations represent causal relationships.
The conflict between the different theoretical approaches to metropolitan area governance has been unresolved because the existing empirical studies on the subject have been largely inconclusive or contradictory. A major reason behind the lack of conclusive evidence is the absence of good quality and cross-country comparative data on metropolitan area governance structures. Large metropolitan area datasets usually contain only basic variables, such as population size and number of administrative units and layers. Detailed information on the particular forms and characteristics of metropolitan area governance structures are often only available on selected individual metropolitan areas collected through case studies. However, this data is not representative because metropolitan areas with interesting or more innovative governance structures tend to get analysed more often than others. Furthermore, different case studies usually use different categories. This makes it difficult to combine information from case studies in one data set.
This project aims to bridge the gap between those two approaches and provide a new perspective at metropolitan area governance. It has two primary goals. First of all, it collects representative data on governance arrangements in nearly all metropolitan areas in the OECD. While not as detailed as data from case studies, the level of detail goes significantly beyond the level of detail of existing governance data sets of metropolitan areas. Secondly, the newly collected data is used for two purposes. It provides the foundation for a representative description of the governance structures in metropolitan areas across the OECD. The descriptive part covers both qualitative and quantitative aspects. Finally, correlations between the governance structures and economic and social outcomes in metropolitan areas are analysed. While this part cannot claim to identify causal relationships, it is indicative that such relations might exist.
The study is part of a wider OECD research project on "Urban Trends and Governance". Within the research project, two further working papers on metropolitan area governance exist. Kim, Schumann and Ahrend (2014) focuses on a qualitative characterisation of typical metropolitan area governance arrangements and Ahrend and Schumann (2014) provides a description of metropolitan governance structures by country.
The remaining part of the paper is structured as follows; the first section provides an overview of the data gathering process and gives a definition of the types of organisations that are considered in this study. Descriptive statistics of the types of governance arrangements are presented in the following section. The subsequent section uses the collected data to show correlations between governance characteristics and outcome variables, and a final section concludes.

Sampling Strategy and Methodology
The OECD defines metropolitan areas as functional urban areas with at least 500,000 inhabitants. A functional urban area is an urban agglomeration with a continuously built-up urban core and surrounding areas. The limits of the functional urban area are determined by the share of the inhabitants that commute from the surrounding areas into the urban core (see OECD, 2012, for a description of the methodology used to define functional urban areas). In total, 275 urban agglomerations within the OECD have been classified as metropolitan areas.
The project aims at collecting information for all metropolitan areas. In order to achieve the objective of complete coverage, data has been systematically collected from publicly available sources on the internet. There are several advantages and disadvantages related to this form of data collection. The foremost advantage of this method is that it can overcome shortcomings of traditional direct survey instruments that have typical response rates well below the rate that is required to achieve representative coverage of metropolitan areas in all countries. However, there are some notable drawbacks related to the chosen method of data collection. Firstly, the amount of available information is limited and varies across cases. Factual information is often available, but qualitative information tends to be absent. For example, there is generally information on the fields of work of a governance body on its website. However, it is rarely possible to obtain any precise information on the degree of involvement in a particular field of work. Similarly, there is often information on the formal powers of a governance body, but hardly any information on whether these powers allow for an effective decision making process. Secondly, the quality of the data is only as good as the quality of the information published online. While only sources that are generally considered as trustworthy have been used, it is difficult to confirm their accuracy in every case. In this context, the biggest concern relates to outdated rather than outright wrong information, as the frequency of updates on many websites is not always stated and may not be high in some cases.
Lastly, there is the possibility that organisations exist that do not have an online presence or that could not be found during the data gathering process. While nowadays it appears unlikely that an organisation, which is involved in metropolitan area governance, leaves no traces online, it cannot be ruled out. Similarly, it cannot be ruled out either that an organisation works on particular issues that are not mentioned in any document that is published online.
The study focuses on organisationscalled metropolitan governance bodiesthat have been put in place to provide metropolitan area governance. Box 1 provides a definition of them. It does not cover other forms of cooperation between municipalities in a metropolitan area that occurs on a caseby-case basis or through term contracts, nor any form of informal cooperation, such as handshake agreements between mayors. Both types of cooperation can play consequential roles in metropolitan area governance and their omission is no judgement of importance, but simply due to feasibility issues in the data collection.

Box 1. Definition of Metropolitan Area Governance Body
The nature of organisations that are supposed to help with the governance of metropolitan areas differs widely across and within countries. Any definition of metropolitan area governance body has to consider two aspects. On the one hand, it needs to be broad enough to capture the variety of organisations that exist across the OECD and should include local solutions that differ from mainstream approaches to metropolitan area governance. On the other hand, the definition needs to be narrow enough to remain meaningful. It has to distinguish metropolitan area governance from other forms of governance that exist throughout the OECD. As other forms of multi-level governance, metropolitan area governance is characterised by the vertical and horizontal interactions of different actors that can take on many different forms (see Hooghe and Marks 2001 and subsequent literature). In order to take these interactions into account, a definition of metropolitan area governance must contain multiple criteria that define governance bodies along the different dimensions of these interactions.
In order to be classified as a metropolitan area governance body for the purpose of this project, an organisation has to satisfy the four criteria below:

Geographical scope: The organisation must cover the central city and a large share of the remaining parts of the metropolitan area. If its geographical scope extends beyond the metropolitan area, the metropolitan area must constitute the predominant part of its sphere of responsibility.
The geographical focus of an organisation has to lie on the metropolitan area in order to be considered its governance body. In particular, the central city as well as the surrounding areas have to be represented in it. Coordination between the central city and surrounding areas is a crucial aspect of metropolitan governance. Any organisation that is considered a metropolitan area governance body must work on such coordination issues and therefore cover the urban core and surrounding areas.
In practice, it rarely happens that the geographical extent of a potential governance body coincides perfectly with the geographical extent of the metropolitan area. Furthermore, in many countries there is no official definition of metropolitan area. Therefore, the geographical scope can be somewhat different from the metropolitan area as long as it shows a clear focus on the metropolitan area.

Involved actors: National or sub-national governments must be dominant actors within the organisation or, alternatively, the organisation itself has to have the status of a sub-national government.
National and sub-national governments are the most important actors in metropolitan area governance. This has to be reflected in the composition of the governance body, which must be predominantly composed of representatives of such governments. While governments are the most important actors in metropolitan area governance, they are not the only ones. Therefore, the condition does not rule out that other actors, such as the business community or representatives of civil society are represented on the governance body. In some countries, organisations that were created explicitly for the purpose of metropolitan area governance have the status of a full local government. These organisations typically present the most integrated approach to metropolitan area governance and are also considered governance bodies.

Thematic focus: The organisation must primarily deal with issues that are directly and predominantly relevant to metropolitan area governance.
Metropolitan area governance concerns a wide range of topics. While it is not possible to provide an exhaustive list, all of them have in common that they concern policy fields that require coordination between parts of the metropolitan area. Often, these are issues where decisions in one part of the metropolitan area have spillover effects on other parts. Direct relevance means that issues should appear more frequently or have a higher relevance in metropolitan areas than in other areas. In practice, the condition implies that most organisations, which cover a metropolitan area without being created as metropolitan area governance bodies are not considered as such. For example, the jurisdictions of some sub-national governments coincide with metropolitan areas. Usually, such sub-national governments were not created with metropolitan area governance in mind and do not focus on it. Instead, they fulfil functions that are similar to other sub-national governments of the same level that to not cover metropolitan areas. Therefore, they are not considered governance bodies as long as they have not received particular powers or responsibilities that strengthen their role as metropolitan area as governance body.

Thematic width: An organisation must have a mandate that allows it to work on more than one issue that is related to metropolitan area governance.
This criterion serves to distinguish metropolitan area governance bodies from single issue bodies and sectoral authorities. Metropolitan area governance always concerns a variety of issues. Furthermore, these issues can rarely be viewed in isolation because they tend to interact with each other. Any organisation that can be considered a metropolitan area governance body must be able to address this complexity. Of course, it is not guaranteed that every organisation, which deals with at least two different issues, addresses the complexities of metropolitan area governance in a meaningful way. Nevertheless, the distinction between organisations that focus on one issue and organisations that focus on many issues is an operationally useful and objective way to identify organisations that work across individual policy fields.
The definition of metropolitan governance body makes no reference to the powers of an organisation. This implies that the spectrum of organisations that are considered governance bodies ranges from purely consultative ones to those that have extensive legal powers. A definition that aims to capture governance bodies across a wide range of countries cannot be more restrictive along this dimension because it needs to be applicable to very different systems of local government. A further implication of the definition above is that metropolitan areas with a governance body are not necessarily less polycentric in their governance structure than those without. Some types of governance bodies certainly reduce the degree of administrative polycentricity by centralising many powers within a single authority. Others, however, can increase administrative polycentricity by adding centres of decision making without substantially reducing the influence of others.
No matter whether governance bodies increase or decrease the administrative complexity of a metropolitan area, it is a priori unclear whether they improve or worsen its governance. On the one hand, they might improve the information exchange within the metropolitan area, overcome coordination problems among municipalities and create economies of scale. On the other hand, they can increase the size of the public administration beyond its efficient level, reduce the accountability of decision makers and increase legislative complexity by adding another level of local government.

The Data Gathering Process
The questionnaires (see Appendix 3) regarding the governance structures of each metropolitan area were completed with information gathered from online sources. Typically, these were websites of governance bodies, in particular the "About Us" and "FAQ" sections. Often, the statutes and regulations, annual reports and balance sheets provided useful data sources, too. In some cases, additional information has been collected from websites of other local governments, from academic publications and from newspaper articles.
The collected information was double-checked for validity and subsequently coded and transcribed into a spreadsheet. In an effort to further validate the information, the spreadsheets were sent out to representatives from OECD member countries responsible for regional and urban development for verification during the last quarter of 2013. In general, the data reflects the situation as it was presented in the sources during the time of the data collection. 4

DESCRIPTIVE STATISTICS
The data covers 263 metropolitan areas. Thus, more than 90 per cent of all metropolitan areas in OECD countries (as identified by the methodology described in OECD 2012) are included in the data. 5 The Metro Areas covered are listed in Table 1. The average population per metropolitan area is slightly below 2 million, and the median population is slightly above 1 million people. The smallest metropolitan areas in the sample have a number of inhabitants just around the threshold of half a million, whereas Tokyo, the largest one, has almost 35 million inhabitants. For each of the 263 metropolitan areas in the sample, information on 29 variables has been collected. A description of the variables used can be found in the Appendix to this paper.

4.
In individual cases, it cannot be ruled out that in the course of the revision process updates from a later point in time have been included in the data. However, as the entire data collection process occurred within less than a year, this should not be an issue.

5.
Due to data availability issues, the OECD has not yet identified metropolitan areas in Australia and New Zealand. This study nevertheless includes 11 urban agglomerations from those countries that are likely to satisfy the conditions for being classified as metropolitan areas.

How common are Governance Bodies
Metropolitan area governance bodies according to the definition above are common across OECD countries. In total, there are governance bodies in 178 out of the 263 analysed metropolitan areas. This is equivalent to a share of 68 per cent.
Most countries have governance bodies in some but not all of their metropolitan areas. Eleven out of the 21 countries for which data has been collected have metropolitan areas with governance bodies and metropolitan areas without governance bodies. There are seven mostly smaller countries that have governance bodies in all their metropolitan areas and three countries that do not have any dedicated metropolitan area governance bodies. Figure 1 indicates the share of metropolitan areas with governance bodies by country. In recent years, renewed attention has been put on the subject of metropolitan area governance. This is reflected by the age of the existing governance bodies. An analysis of the time when they were created shows a cyclical pattern. A first wave of creations occurred in the 1960s and early 1970s, but the number decreased in the following years and few governance bodies were created throughout the 1980s. In the 1990s the number of creations started to increase again and has peaked in recent years. Figure 2 shows a histogram of the dates of inception of currently existing governance bodies. Note that it uses the dates when government bodies were created in their present form. This can either be their inception date or alternatively the date of the last major reform. 6 Figure 2 does not include the creations of now defunct bodies and therefore should not be interpreted as representing the total 6.
No information is available on whether or not precursor institutions existed and, if yes, when there were founded. Furthermore, no information exists on bodies that have ceased to exist. activity with respect to the creation of governance bodies. However, it appears likely that it gives a good approximation of the actual pattern.

Figure 2. Inception Year of Governance Bodies
When governance bodies exist, they often differ across metropolitan areas even within the same country. This is most obvious with respect to their legal powers. 48 out of the 178 existing governance bodies have the right to impose binding laws or regulations, which equals a share of 27 per cent. Even though less than one third of all governance bodies have the right to impose laws or regulations, bodies with these powers exist in more than half of the analysed countries. Out of the twelve countries in which governance bodies with legal powers exist, six have both governance bodies with and governance bodies without legal powers. This shows the diversity of governance bodies not just across but also within countries.

Fields of Work of Governance Bodies
Governance bodies typically cover several public policy fields that vary greatly across countries. However, three topics stand out as particularly common fields of work. These are regional development, transport and spatial planning. They are among the fields of work of some governance bodies in virtually all countries where such bodies exist.
The most common field of work is regional development, with 81 per cent of all governance bodies covering this field. Governance bodies are considered to be active in this field if their work tries to directly advance the local economy. This can be through initiatives to support certain sectors, to attract specific companies or industries, to affect the structure of the local labour market or to promote the skills of the local population. Similarly, initiatives that aim at the regeneration of particular geographical areas and similar projects are considered part of regional development.
78 per cent of governance bodies work on transportation, the second most common field of work. Bodies active in this field typically work on issues related to public transport, but may alternatively also work on individual transport in general and on roads in particular. The third most common field of work is spatial planning, with 67 per cent of governance bodies working in this area. In total, more than half of all government bodies (91 out of 178) work on all three fields. This combination of fields of work is especially common among European and North American governance bodies.
All other fields of work are significantly less common and are typically found only in certain countries. In order of frequency, these fields are waste disposal (35%), water provision (35%), culture and leisure (29%), tourism (26%), sewerage (26%), and energy (15%). Figure 4 shows the numbers graphically.

. Share of Governance Bodies Active in Selected Fields
When interpreting the number of governance bodies that work in any particular field, it needs to be kept in mind that their powers vary greatly. Accordingly, there are large differences in the actual influence on policies. In the case of spatial planning, for example, some governance bodies merely serve to inform member local-governments of each other's plans, whereas others exert centralised control over the entire planning process in a metropolitan area. In between those extremes, there is a continuum of governance bodies with varying influence on the planning process. All of them are considered to be active in the field although most have rather few formal competencies. It would be desirable to further separate governance bodies depending on the scope of their activities in a particular field of work. However, due to data availability issues, this is not possible at this stage of the project.

Budgets and Staff
Information on their budgets could be found online for 123 out of the 178 governance bodies. Most common are annual budgets of around 10 USD per capita in purchasing power parity terms. The median annual per capita budget is 13.9 USD. However, depending on the tasks of the governance body, budgets can be several orders of magnitude higher. The highest per capita budget in the sample is 3800 USD. Figure 5 shows the distribution of governance bodies' per capita budget (using a logarithmic scale). There are roughly two groups of governance bodies; those with per capita budgets of less than 30 USD and those with per capita budgets of more than 1000 USD. Not surprisingly, per capita budgets are closely associated to the characteristics of the governance body. Those bodies with budgets at the lower end of the scale tend to be voluntary associations of governments that work through soft coordination, whereas those at the upper end of the scale are either full local governments or inter-municipal associations with wide-ranging powers. Information on staff numbers is available online for 95 governance bodies. In 74 cases, governance bodies provide both, staff numbers and a budget. Not surprisingly, the picture for staff numbers corresponds roughly to the one for budgets. In general, there is a somewhat greater variation in per capita staff numbers than in per capita budgets. However, it remains unclear if this represents actual differences in the way governance bodies work or if it is simply due to differences in the way staff numbers are reported (for example, whether part-time and support staff are included in the numbers).

Political Leadership
The precise rules that determine the leadership composition of a governance body are almost always unique. Nevertheless, several regularities appear. Most commonly, bodies are made up of elected officials of the local governments whose territories they cover (see Figure 6). These officials are either appointed to the body by the local governments or obtain a seat by virtue of their office. The latter case is typical for mayors of municipalities. Elected officials from local governments sit on 55 per cent of all governance bodies. In 7 per cent of all cases, governance bodies include representatives of other interest groups or private citizens in addition to appointed officials. Public elections for the leadership of the governance body are held in 11 per cent of all cases. 13 per cent of governance bodies have members that have been selected based on other criteria. In those cases, representatives are typically appointed by national governments or higher levels of sub-national governments. In 14 per cent of the cases, no information could be found online regarding the composition of the governance bodies. Those governance bodies that have public elections for their leadership tend to be are much more likely to have formal rights and responsibilities than those that do not have public elections. In contrast, the governance bodies that include private citizens or representatives of interest groups tend to be the ones with the fewest rights and responsibilities.

Sectoral Authorities for Public Transport
Sectoral authorities and special purpose associations can play important roles for metropolitan area governance. In contrast to governance bodies, sectoral authorities are organisations that have a narrow mandate that focuses on the provision of a particular service. Sectoral authorities vary greatly across metropolitan areas in OECD countries. They might focus on waste disposal, water provision or business promotion. However, public transportation authorities and associations are by far the most common sectoral authorities and often also the ones with the highest budgets.
As it is the case with governance bodies, transport authorities are organised in very different forms across the OECD. On one end of the spectrum, organisations are only responsible for strategic public transportation planning and employ a low double-digit number of staff. On the other end of the spectrum, organisations plan and operate the entire public transportation system of a metropolitan area themselves and employ several tens of thousands of employees. In general, no attempt was made to distinguish between those sectoral authorities, as the amount of available information is fairly limited and the organisational structures tend to be complex and difficult to compare across countries.
An organisation is considered a transport authority if it focuses exclusively on public transport in a metropolitan area and is responsible for its strategic planning, independently from whether or not it has the authority to make strategic decisions. This definition distinguishes transport authorities from 55% 11% 7% 13% 14% Representatives from Municipality Governments Public Elections for Governance Body

Representatives from other Levels of Government
No data other public organisations or private companies that operate the public transport systems in metropolitan areas, but have little influence on their general design.
Transport authorities exist in 56 per cent of all metropolitan areas. They are more common in metropolitan areas that have governance bodies, but also many metropolitan areas without governance body have a public transport authority. In areas with governance body, they exist in 62 per cent of all cases, compared to 42 per cent in areas without one.

GOVERNANCE AND SOCIO-ECONOMIC CONDITIONS IN METROPOLITAN AREAS
The following sections present estimates from regressions of characteristics of metropolitan areas on the newly collected data on governance structures. The regressions control for country specific effects. Therefore, general differences between countries do not affect the results. Box 2 provides details on the methodology behind the estimations.

Box 2. Estimation Details
All estimates are based on regressions that control for country specific effects by using country dummies. Therefore, the results are not influenced by general differences between countries and the coefficients can be interpreted as the difference between the metropolitan areas of different kinds within the same country. While this procedure is generally preferable in order to avoid having the results biased by differences between countries that are not related to the characteristics of metropolitan areas, it has a small downside. If explanatory variables in a regression do not vary across metropolitan areas within a country, these observations cannot be used to estimate the coefficient of interest. It also implies that metropolitan areas from countries with only a single metropolitan area never affect the coefficient of interest (currently, the only country in the data with one metropolitan area is Ireland). With Transport Authority Without Transport Authority

Box 2. Estimation Details (cont.)
All equations are estimated by linear regression. Unless otherwise mentioned, they have the following general form: , where y represents the outcome variable of interest, α is an intercept, β is the coefficient of interest on the explanatory variable x. Z is a vector of control variables and the corresponding vector of coefficients. C is a set of country-dummies that are used to control for country fixed-effects and ε the error term. Different metropolitan areas are indicated by subscript i. The estimated correlations show a relation between the structure of metropolitan governance and outcomes in the metropolitan area that might be considered suggestive of a causal relationship. 7 It is rare that the OECD definition of a metropolitan area corresponds precisely to its local definition. Furthermore, governance bodies often cover only parts of metropolitan areas or extend their work beyond their borders. Therefore, outcome variables almost never correspond exactly to the area in which the governance body is active. This is a form of measurement error that affects the right-hand-side variables in the estimation and can lead to an underestimation of the true correlations.

Governance Bodies, Metropolitan Area Population and per capita GDP
Per capita GDP and the existence of metropolitan area governance bodies are positively correlated. Those metropolitan areas that have a governance body have a per capita GDP that is on average 7 per cent higher than that of metropolitan areas without governance body. Controlling for log-population levels reduces the point estimate to 4 per cent. When adding further control variables, such as population density, the number of urban cores and the degree of local government fragmentation, the point estimate fluctuates between 4 and 7 per cent and tends to be marginally significant. In contrast, there is a strong and stable correlation between the size of a metropolitan area and the existence of a governance body. Metropolitan areas that have governance bodies are approximately one third larger than those that do not. 8 Metropolitan areas with governance bodies are not only larger than those without, they have also been growing faster between 2000 and 2010. The yearly population growth rate in the former group of metropolitan areas was 0.3 percentage points higher than the growth rate of the latter group. 9

7.
Before interpreting the results as a causal relationship, the usual caveats of omitted variables and reverse causality should be carefully considered, as they are affecting the results to an unknown degree. For statistical inference, heteroskedasticity-robust standard errors have been used. These do not take unobserved correlations between the error terms into account and therefore might underestimate true standard errors. However, due to the relatively small number of observations, the usual remedies, such as clustering errors on the country-level to account for the most common sources of correlation between errors, are not feasible. See Angrist and Pischke (2008) for a detailed treatment of these issues.

8.
The effect is precisely estimated and robust across many different specifications. It furthermore holds within countries as well as between countries.

9.
The estimate is statistically significant at the 99% confidence level and robust against the inclusion of control variables, such as population levels in 2000, population density per capita GDP in 2000.

Population Density and Urban Sprawl
The existence of a governance body is strongly negatively correlated with the development of urban sprawl. Metropolitan areas with a governance bodies experienced a small decline in urban sprawl between 2000 and 2006, whereas those without a governance body experienced an increase in sprawl. 10 Sprawl in this context is defined by the change of the population density of the build-up area. 11 Figure 8 shows this relationship.

Figure 8. Change in Sprawl
Mirroring the results for urban sprawl are the estimates for the growth of surface of the urban agglomeration. In metropolitan areas with governance bodies it grew roughly 0.2 percentage points less than in metropolitan areas without governance bodies. Once one controls for the population growth rate, this difference increases to 0.34 percentage points. As the average growth rate of urban surface in metropolitan areas without governing bodies is 0.83 per cent, the difference is substantial. While the existence of a governance body is not significantly correlated with population density, population density is approximately 20 per cent higher in metropolitan areas, which have governance bodies that work on spatial planning issues. 12

10.
The result is statistically significant at the 99 per cent level and highly robust to the inclusion of control variables such as per capita GDP, per capita GDP growth and population density.

11.
See http://measuringurban.oecd.org for an exact definition. The recent literature on urban form defines sprawl as a multidimensional phenomenon of which population density is only one aspect. These other dimensions of sprawl are not considered in this analysis.

12.
This does not necessarily imply a causal relationship. As mentioned above, a decline in urban sprawl between 2000 and 2006 is associated with the existence of a governance body. However, it is also associated with more frequent creations of governance bodies between 2007 and 2013. As those bodies could not affect the development of sprawl before they were founded, there must be other unobserved factors that are responsible for both developments. A potential interpretation of this result is related to the actual cooperation of local governments within a metropolitan area. It might not be the With Governance Body Without Governance Body

Fields of Work
No statistically significant correlations can be found between the fields of work of governance bodies and outcome variables. In other words, after controlling for country-specific effects, whether or not a governance body works in a particular field is not correlated to any observable characteristics of the metropolitan area. 13

Public Transport Authorities
The existence of a separate public transport authorities is positively correlated to desirable outcomes. For example, air pollution is significantly lower in metropolitan areas with transport authorities. After controlling for per capita GDP population levels and population density, metropolitan areas that have public transport authorities report levels of particulate matter (PM2.5) in the air that are approximately 9 per cent lower than those of metropolitan areas without transport authorities.
An important measure of the quality of public transportation is the satisfaction of citizens with the public transport systems in their cities. Data on the perception of public transport systems is available for 35 cities in 11 European countries. 14 It reveals that the share of satisfied citizens is 13 percentage points higher in those metropolitan areas that have a public transport authority than in those metropolitan areas that do not have one. 15 Figure 9 shows this relationship. existence of a governance body per se that leads to reduced sprawl, but better cooperation of local governments in general. Local governments that cooperate well might be more successful in combating sprawl and more likely to create governance bodies, thus causing the correlation between the two variables.

13.
It remains unclear whether this is due to the precision of the estimate, because of the quality of the data or because there is in fact no such correlation. A possible explanation lies in the fact that the data does not account for the actual powers that a governance body has in a particular. It might be that a binary distinction whether or not a governance body is active in a field is not precise enough to measure the actual consequences of its work.
14. See Feldmann (2008) for details on the methodology of the data collection.

15.
In contrast to the previous estimations, it is in practice impossible to control for country specific effects with such limited data. The low number of observations makes it unlikely that precise estimates can be achieved when many control variables are included in the regression. More importantly, in many countries there is no variation in whether or not a transport authority exists in the metropolitan areas for which data is available. As mentioned above, those observations would be lost for the estimation of the coefficient of interest. However, it is arguably less important to control for countries specific effects when only European countries are analysed, as they tend to be fairly similar. Therefore, a simple regression of the share of citizens that are satisfied with their public transport system on the existence of a public transport authority estimated.

Transparency and Availability of Information
Increased transparency is likely to promote the quality of public administration through several mechanisms. It promotes accountability, reduces corruption and generates trust in the administration. While transparency of public administrations is a complex concept that has many dimensions, a simple indicator of the level of transparency is the amount of information that can be found online. Two important pieces of information in the context of this study are the yearly budget of the governance body and its date of creation. It is obvious that the availability of this information in itself is not very likely to have large effects on outcomes in a metropolitan area. However, it should be considered a proxy for the general openness of the governance body.
Although both pieces of information must exist in general for all governance bodies, only 108 out of 178 governance bodies make both of them available online. It turns out that metropolitan areas, in which governance bodies provide this information, have a six per cent higher per capita GDP than those metropolitan areas where governance bodies do not provide this information. As mentioned earlier, it is important to note that this result should not be considered a causal estimate. At most it can be considered indicative that greater transparency has positive effects on important outcomes.

Conclusion
This study provides the first representative overview of governance structures in virtually all of the metropolitan areas in OECD countries. Using online research as method of data collection, it presents data on metropolitan area governance bodiesorganisations that have been put in place to support metropolitan area governance. These organisations contribute to the governance of metropolitan areas either directly through their own legislative or regulatory powers or indirectly by supporting the cooperation among other local governments within the metropolitan area.
The study first of all shows that in most metropolitan areas organised approaches to metropolitan area governance exist. However, in a majority of the cases, the organisations that have been created in With Transport Authority Without Transport Authority % of Citiziens Satisfied with Public Transport System these metropolitan areas have few formal powers and serve mostly as an institutionalised forum for policy makers to exchange and coordinate policies. This is also visible in the budgets of these organisations, which tend to be small on a per capita basis. In terms of the theoretical debate about optimal governance structures mentioned at the beginning of this paper, it is not clear if these weak governance bodies are expressions of centralised or polycentric structures. On the one hand, they could be considered the weakest form of a centralised decision structure and a first step towards a more consolidated form of governance. On the other hand, it is possible to interpret these organisations as a result of the informal connection that public choice theorists such E. Ostrom (2010) predict in fragmented metropolitan areas.
The data clearly confirms the importance of three policy fields for metropolitan governance; regional development, transportation and spatial planning. Judging by the attention they receive from policy makers, they can be considered the defining issues of metropolitan governance. Especially transportation has a special status in metropolitan governance that is also documented by the fact that more than half of all metropolitan areas have dedicated transport authorities. Metropolitan wide transportation authorities are common also in countries that have otherwise no tradition of sectoral authorities that cover the territory of several municipalities.
The analysis shows that metropolitan areas with governance bodies perform better on several dimensions. They are denser, have higher per capita GDPs and attract more people. In addition, there is a clear correlation between the existence of transport authorities and the satisfaction of citizens with public transport services, as well as environmental outcomes. These results can be seen as indicative of the positive effects of governance bodies. 16 To sum up, the study provides a new way to look at metropolitan area governance. The collected data gives the first representative overview of the governance structures in more than 90 per cent of all metropolitan areas in OECD countries. It allows for quantitative descriptions and analyses of the existing governance structures. Thereby, it provides a novel perspective on metropolitan area governance and gives researchers new tools for studying this important subject.

16.
Nevertheless, it is important to emphasise the well-known fact that correlation does not necessarily imply causation. While the reported results are robust to the inclusion of some of the most important control variables it is likely that there are other omitted factors that could not have been taken into account. It remains unclear to what degree they bias the estimates.

Metropolitan Area
Name of the metropolitan area as it appears in the OECD Metropolitan Areas Database

Primary City
Name of the central city of the metropolitan area

Number of MA Jurisdictions
The total number of all local governments within the metropolitan area. This includes every local government that lies partially or completely within the metropolitan area, but it excludes those governments that completely encompass the metropolitan area. For example, if a province completely covers the metropolitan area, the province would not be included in this figure. However, if the metropolitan area falls partially into two provinces, both of them would be included.

MA has Governance Body
Does a metropolitan area governance body exist in the metropolitan area? See above for the definition of metropolitan area governance body that is used for this project.

Body Name
Name of the governance body

Inception Year
The year the body was created in its current form. This can either be the year of its foundation or the year of the last major reform of the body. A major reform is any reform that changed the structure, the membership or the legal powers of a governance body in a substantial way.

Legislative/Regulatory Powers This variable indicates if a Governance
Body has legislative or delegated legislative powers.

Number of Representatives
The total number of representatives that serve on the governance body

Staff of the Governance Body
The total number of people working primarily for the governance body. This number does not include employees of other local governments within the metropolitan area.

Governance Body Budget
Budget of the governance body in local currency

Transportation
The governance body works on transportation issues *

Spatial Planning
The governance body works on spatial planning issues*

Regional Development
The governance body works on regional development (i.e. tries to directly influence the local economy). This can be through initiatives to support certain sectors, to attract specific companies or industries, to affect the structure of the local labour market or to promote the skills in the local population. Furthermore, initiatives that aim at the regeneration of particular areas and similar projects are considered part of regional development.*

Waste Disposal
The governance body works on waste disposal issues*

Water Provision
The governance body works on water provision issues*

Sewerage
The governance body works on sewerage issues*

Energy
The governance body works on issues related to energy provision*

Education
The governance body works on education issues*

Tourism
The governance body works on tourism issues*

Culture and Leisure
The governance body works culture and leisure issues*

Health
The governance body works on public health issues*

Transportation Authority
Name of the sectoral transportation authority

Information on the city centre
What is the name of the primary city (potentially identical with city/MA name provided above)? NB: If there is more than one city centre, please list all of them and include this information for each of them in the follow-up questions on city centre. For example for Manchester it would be Manchester. The primary city of the MA of San Francisco has Oakland and San Jose as other city centres. Source:

Information on MA governance
How many jurisdictions/municipalities exist in the entire MA?
NB: Simply insert total number of jurisdictions/ municipalities independent of type. This question corresponds to jurisdictions in the form of cities, counties, communes etc.not in form of single purpose authorities such as a transport authority. Source: